Does Increasing Mother’s Schooling Reduce Fertility and Increase Children’s Education: Evidence from a Natural Experiment on Arabs in Israel Victor Lavy and Alexander Zablotsky Preliminary and Incomplete June 17, 2010 Abstract We study in this paper the causal relationship between women’s education and fertility and the transmission of human capital from mothers to children. We base our evidence on a natural experiment that increased sharply the education of affected cohorts of children as a result of the de facto revocation in October 1963 of the Military Government of Arabs in Israel which immediately enabled a large part of the Arab population to regain access to schooling institutions. The Military Government which was in effect from 1948 imposed severe restrictions on movement and travel and therefore disrupted sharply access to schooling of residents in localities that lacked education institutions. Regaining access to schooling increased female years of schooling by 1.42 for women age 4-9 in 1964 and by 0.75 for women age 9-13 at that time. These gains reflect an increase of 16.4 and 6.1 percent increase in the probability of completing primary school for the two affected cohorts, respectively. These very large effects on schooling levels led to a sharp decline in completed fertility, 0.6 children for the younger affected cohorts and of 0.4 children for the older cohorts. The implied 2SLS estimates show that an increase in one year of maternal schooling caused a decline in fertility of 0.5 children. Additional evidence we present suggest that labor force participation, age when married, marriage and divorce rates as well as spouse education and earnings did not play a role in this fertility decline. In the second part of the paper we show that the increase in mother schooling led also to an increase in the education of children in face of a decline in their number. 1. Introduction In the economic model of fertility (Becker, 1960, Mincer, 1963), education increases the opportunity cost of women’s time leading them to have fewer children, but also raises a woman’s permanent income through earnings, tilting her optimal fertility choices toward higher quality. (Becker and Lewis 1973, Willis 1973). Second, under positive assortative mating, a woman’s education is causally connected to her mate’s education (Behrman and Rosenzweig 2002), so the effect of education on household permanent income is augmented through a multiplier effect. However, there are societies that experienced fertility transition without these economic forces playing a major role. For example, during half century the total fertility rate of Moslem women in Israel fell sharply, from over 9.8 children in the mid 1950’s to 3.9 children in 2008.1 During the same period, Israeli Arab women’s average years of schooling increased by more than three folds, from 3 years in 1951 to over 10 years in 2008, but this change barely impacted their labor force participation and employment during the same period reaching only 15 percent in 2000 and 18 percent in 20092. However, the education increase could have impacted Arab's women fertility through other channels. First, education may improve an individual’s knowledge of, and ability to process information regarding fertility options and healthy pregnancy behaviors (Grossman 1972). Second, education may equip girls with a better ability to process information, potentially increasing knowledge of contraception options (Rosenzweig and Schultz, 1989, Thomas, Strauss, and Henriques, 1991). Education may also improve the wife’s bargaining power inside the marriage (Thomas, 1990). Moav (2005) suggests that educated mothers may have a technical advantage in producing educated children, tilting the trade off from quantity to quality children. However, there little evidence of the importance of these channels in the absence of a meaningful increase in women's employment and opportunity cost of time. The objective of this paper is to study the role that female education played in reducing fertility while tradition still kept women from the market place. In particular, we present evidence that indicates that the strong negative relationship between women's fertility and education reflects a causal effect and show that potential mechanisms such as women's labor force participation, age when married, marriage, and divorce rates did not play a major role in this fertility decline. The impact of women's education is still very large after accounting for spouse's education and family income. In the last section of the paper we show that the increase in mother schooling led also to an increase in the education of children in face of a decline in their number. 1 Central Bureau of Statistics’s website, online tables and figures. 2 Central Bureau of Statistics (2002), State of Israel Prime Minister’s Office and Yashiv and Kasir (2009). 1 We base the evidence presented in this paper on a natural experiment that increased sharply the education of affected cohorts of children as a result of the de facto revocation in October 1963 of the Military Government of Arabs in Israel which immediately enabled a large part of the Arab population to regain access to schooling institutions. The Military Government was in effect from 1948 to 1966 over some geographical areas of Israel having large Arab populations, primarily the South (Negev), North (Galilee), and the central (Triangle). The residents of these areas were subject to a number of measures that placed tight controls on all aspects of life of Arabs in Israel. These measures included severe restrictions on movement, and permits from the military governor had to be procured to travel more than a given distance from a person's registered place of residence. Although the Military Government was officially for geographical areas and not people, its restrictions were enforced only on the Arab residents of these areas. The travel restrictions were revoked in October 1963 following the resignation in June of that year of the Prime Minister, David Ben Gurion, who together with his ruling labor party strongly supported the continuation of the Military Government. The change was also a response to the mounting pressure from the Israeli public and from many political parties, including the Heirut right wing party, to annul the Military Government of Israeli Arabs. This effort led in 1966 to the complete cancelation of The Military Government and the Arab citizens of Israel were, theoretically if not always in practice, granted the same rights as Jewish citizens. The Military Government restricted de facto access to schools for children in localities and villages in which there were no primary or secondary schools while not affecting access in localities in the Military Government regions that had schools already. It therefore created two zones in Arab populated areas, one in which school attendance required some travel which then became difficult or even impossible and one where schooling access was not interrupted at all. In the latter group we distinguish between Arab localities that were under the Military Government and the Arab population living in predominantly Jewish cities that were also placed initially (1948) under military but later were exempted de facto from some of the restrictions. The change in the end of 1963 could benefit children that were not too old to attend primary school or children who completed primary schooling and could now enrol in secondary schooling. Therefore, the exposure of an individual to this ―treatment‖ was determined both by her location and by her year of birth. After controlling for locality and year of birth fixed effects, we use the interaction between a dummy variable indicating the age of the individual in 1964 and whether or not her locality was part of the Military Government zone and had no schools as an exogenous variable, and as an instrument for an individual's education. Similar identification strategies were used to estimate the effect of school quality on returns to education (Card and 2 Krueger, 1992), the effect of change of languish of instruction on the return to schooling (Angrist and Lavy, 1997), the effect of college education on earnings, (Card and Lemieux, 1998), the effect of school construction on education and earnings, (Duflo, 2000) and the effect of school competition on pupil’s academic achievement (Lavy, 2010). We allow the affected cohorts to include children from age 2 to age 13 in 1964 while older cohorts are used as controlled experiments. We use data from the 1983 and 1995 census. In 1983 the affected cohorts were 24 to 35 years old which permits studying the effect of education on early age fertility. In 1995 the affected control cohorts were already 40 to 53 years old, allowing estimating the effect of education on completed fertility. The evidence we present in the paper suggests that regaining access to schooling starting 1964 increased female years of schooling by 1.42 for the women at age 4-9 in 1964 and by 0.75 for women at age 9-13 at that time. The gain for the respective cohorts of men is 0.85 and 0.45 years of schooling. These educational gains are associated with a 16.4 percent increase in the probability that a woman in the young affected cohorts completed primary school and a 6.1 percent increase in the same probability for the older affected cohorts. These very large effects on schooling levels led to a sharp decline in fertility of 0.6 children for the younger affected cohorts and of 0.40 children for the older cohorts. The implied 2SLS estimates show that an increase in one year of maternal schooling caused a decline in fertility of 0.5 children.